Ground segmentation on 3D point clouds is critical for applications such as robotic localization, path planning, and obstacle prediction, where accurate ground segmentation serves as a foundational preprocessing step requiring both computational efficiency and segmentation accuracy. This paper introduces GenericGrid, a versatile ground segmentation method designed to adapt to diverse sensor configurations while delivering robust performance in complex environments. By leveraging 2D height maps and slope maps, the proposed approach effectively addresses terrain modeling and ground segmentation challenges. Evaluations on the SemanticKITTI and Livox Simu-dataset v1.0 demonstrate that GenericGrid achieves state-of-the-art performance, attaining a 94.78% average Intersection-over-Union (IoU) for ground segmentation with near-real-time processing at 120 Hz. Notably, the method exhibits exceptional compatibility with onboard LiDAR systems, underscoring its practicality and reliability for real-world robotic perception tasks.